Skip to main content

Enterprise

AI infrastructure your platform team can sign off

Everything in Team, plus the controls security, compliance, and procurement need before a gateway can carry production traffic. SSO, audit logs, SLAs, dedicated regions — and a CSM who picks up the phone.

SSO + SAML / OIDC

Okta, Azure AD, Google Workspace, JumpCloud — bring your IdP. SCIM 2.0 user + group provisioning included.

Audit logging + SIEM export

Every routing decision, key use, role change, and admin action is logged and exportable to Splunk, Datadog, or your S3 of choice.

99.99% uptime SLA

Penalty-backed availability targets with quarterly credits and a named CSM who carries the pager during incident response.

Dedicated support

Slack Connect or Teams channel, a one-hour first-response on Sev-1, and a quarterly architecture review with our gateway team.

On-prem + VPC option

Air-gapped or self-hosted deployments for regulated workloads. Helm chart + Terraform module, with quarterly upgrades.

Regional + sovereign routing

Constrain every request to specific regions (EU, US, AU) to satisfy GDPR, HIPAA, or data-residency contracts.

Team vs Enterprise

What you pick up moving from a self-serve plan to a signed contract.

FeatureTeamEnterprise
OpenAI-compatible API
Smart routing + semantic cache
BYOK + provider keys
Usage analytics + cost attribution
SSO / SAML / OIDC
SCIM 2.0 provisioning
Audit logs + SIEM export
Dedicated routing region
99.99% uptime SLA
Dedicated CSM + Slack Connect
On-prem / VPC deployment
Custom MSA + DPA

FAQ

How long does an enterprise rollout take?+

Typical engagement is 2–4 weeks from kickoff to production traffic. SSO is configured day one; audit-log export and SCIM in the first week; full rollout in the third.

Do you sign our paper, or only yours?+

We sign customer MSAs, DPAs, and BAAs. We also have our own templates ready if you don't want to drive the negotiation.

Can we self-host?+

Yes. The on-prem build is the same gateway binary, deployed via our Helm chart or Terraform module. Quarterly upgrade cadence, optional managed-update channel.

What about model fine-tunes / private models?+

Bring your own fine-tunes from supported providers, or point at a private inference endpoint via our adapter SDK. The router treats it like any other provider.

Get a 30-minute architecture call

Tell us about your traffic shape, providers you want covered, and any compliance constraints. We’ll show you what a rollout looks like end-to-end and follow up with a concrete proposal.

Schedule a call